Deep learning for segmentation of colorectal carcinomas on endoscopic ultrasound.

IF 2.7 3区医学 Q2 GASTROENTEROLOGY & HEPATOLOGY Techniques in Coloproctology Pub Date : 2024-12-13 DOI:10.1007/s10151-024-03056-5

F van den Noort, F Ter Borg, A Guitink, J Faber, J M Wolterink

{"title":"Deep learning for segmentation of colorectal carcinomas on endoscopic ultrasound.","authors":"F van den Noort, F Ter Borg, A Guitink, J Faber, J M Wolterink","doi":"10.1007/s10151-024-03056-5","DOIUrl":null,"url":null,"abstract":"Background: Bowel-preserving local resection of early rectal cancer is less successful if the tumor infiltrates the muscularis propria as opposed to submucosal infiltration only. Magnetic resonance imaging currently lacks the spatial resolution to provide a reliable estimation of the infiltration depth. Endoscopic ultrasound (EUS) has better resolution, but its interpretation is investigator dependent. We hypothesize that automated image segmentation of EUS could be a way to standardize EUS interpretation.Methods: EUS media and outcome data were collected prospectively. Based on 373 expert manual segmentations, a convolutional neural network was developed to perform segmentation of the submucosa, muscularis propria, and tumors. The mean surface distance (MSD), maximal distance between segmentations (Hausdorff distance; HDD), and overlap (Dice similarity index; DSI) were calculated.Results: The median MSD and HDD values were 3.2 and 17.7 pixels for the tumor, 3.4 and 24.7 pixels for the submucosa, and 2.6 and 20.0 pixels for the muscularis propria, respectively. The median DSI values for the tumor, submucosa, and muscularis propria were 0.82, 0.57, and 0.59, respectively. These values reflect good agreement between manual and deep learning segmentation.Conclusions: This study found encouraging results of using automated analysis of EUS images of early rectal cancer, supporting further exploration in clinical practice.","PeriodicalId":51192,"journal":{"name":"Techniques in Coloproctology","volume":"29 1","pages":"20"},"PeriodicalIF":2.7000,"publicationDate":"2024-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Techniques in Coloproctology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1007/s10151-024-03056-5","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"GASTROENTEROLOGY & HEPATOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Background: Bowel-preserving local resection of early rectal cancer is less successful if the tumor infiltrates the muscularis propria as opposed to submucosal infiltration only. Magnetic resonance imaging currently lacks the spatial resolution to provide a reliable estimation of the infiltration depth. Endoscopic ultrasound (EUS) has better resolution, but its interpretation is investigator dependent. We hypothesize that automated image segmentation of EUS could be a way to standardize EUS interpretation.

Methods: EUS media and outcome data were collected prospectively. Based on 373 expert manual segmentations, a convolutional neural network was developed to perform segmentation of the submucosa, muscularis propria, and tumors. The mean surface distance (MSD), maximal distance between segmentations (Hausdorff distance; HDD), and overlap (Dice similarity index; DSI) were calculated.

Results: The median MSD and HDD values were 3.2 and 17.7 pixels for the tumor, 3.4 and 24.7 pixels for the submucosa, and 2.6 and 20.0 pixels for the muscularis propria, respectively. The median DSI values for the tumor, submucosa, and muscularis propria were 0.82, 0.57, and 0.59, respectively. These values reflect good agreement between manual and deep learning segmentation.

Conclusions: This study found encouraging results of using automated analysis of EUS images of early rectal cancer, supporting further exploration in clinical practice.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

深度学习在内窥镜超声波上分割结直肠癌。

背景：如果肿瘤浸润固有肌，而非仅浸润粘膜下层，早期直肠癌保肠局部切除术的成功率较低。目前，磁共振成像缺乏空间分辨率，无法对浸润深度做出可靠的估计。内窥镜超声（EUS）具有更好的分辨率，但其解释取决于研究人员。我们假设，对 EUS 进行自动图像分割是实现 EUS 解释标准化的一种方法：方法：前瞻性地收集了 EUS 介质和结果数据。在 373 次专家手动分割的基础上，我们开发了一种卷积神经网络，用于对粘膜下层、固有肌和肿瘤进行分割。计算了平均表面距离（MSD）、分割间最大距离（Hausdorff距离；HDD）和重叠度（Dice相似性指数；DSI）：肿瘤的 MSD 和 HDD 中值分别为 3.2 和 17.7 像素，粘膜下层分别为 3.4 和 24.7 像素，固有肌分别为 2.6 和 20.0 像素。肿瘤、黏膜下层和固有肌的 DSI 中值分别为 0.82、0.57 和 0.59。这些数值反映了人工分割与深度学习分割之间的良好一致性：本研究发现，对早期直肠癌的 EUS 图像进行自动分析取得了令人鼓舞的结果，支持在临床实践中进一步探索。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊

Techniques in Coloproctology GASTROENTEROLOGY & HEPATOLOGY-SURGERY

CiteScore

5.30

自引率

9.10%

发文量

176

审稿时长

1 months

期刊介绍： Techniques in Coloproctology is an international journal fully devoted to diagnostic and operative procedures carried out in the management of colorectal diseases. Imaging, clinical physiology, laparoscopy, open abdominal surgery and proctoperineology are the main topics covered by the journal. Reviews, original articles, technical notes and short communications with many detailed illustrations render this publication indispensable for coloproctologists and related specialists. Both surgeons and gastroenterologists are represented on the distinguished Editorial Board, together with pathologists, radiologists and basic scientists from all over the world. The journal is strongly recommended to those who wish to be updated on recent developments in the field, and improve the standards of their work. Manuscripts submitted for publication must contain a statement to the effect that all human studies have been reviewed by the appropriate ethics committee and have therefore been performed in accordance with the ethical standards laid down in an appropriate version of the 1965 Declaration of Helsinki. It should also be stated clearly in the text that all persons gave their informed consent prior to their inclusion in the study. Details that might disclose the identity of the subjects under study should be omitted. Reports of animal experiments must state that the Principles of Laboratory Animal Care (NIH publication no. 86-23 revised 1985) were followed as were applicable national laws (e.g. the current version of the German Law on the Protection of Animals). The Editor-in-Chief reserves the right to reject manuscripts that do not comply with the above-mentioned requirements. Authors will be held responsible for false statements or for failure to fulfill such requirements.